Big Data Statistical Learning Improves Survival Prediction For Head And Neck Cancer Patients

Abstract

Immunotherapy has emerged as a novel treatment modality for recurrent/metastatic head and neck cancer (R/M HNC) patients, although overall survival remains poor. Given the reliance of immunotherapy on circulating immune cells, we hypothesized that metabolic signatures of nontumoral hematopoietic tissues derived from quantitative analysis of pre-treatment 18-fluorodeoxyglucose-positron emission tomography-computed tomography (PET/CT) could aid in predicting response to immunotherapy. Materials/Methods: We performed a retrospective analysis of 51 R/M HNC patients treated at a single-institution who received salvage immunotherapy following clinical, pathologic, or radiographic failure after upfront management with adjuvant or definitive radiotherapy. The pretreatment PET/CT closest to immunotherapy initiation (limit of 180 days) was analyzed. Metabolic standardized uptake value (SUV) of predefined nontumoral hematopoietic tissues (liver, spleen and bone marrow) were calculated. Patient demographics including clinical stage, primary disease site, p16 status, and hematologic counts were assessed. Univariable and stepwise multivariable (UVA and MVA, respectively) Cox proportional hazard regression analyses were performed for overall survival (OS) and progression-free survival (PFS). Results: Median follow up is 11.1 months (95% CI 9.7-24). Median OS was 10.8 months (95% CI 7.5-18.9), and median PFS was 3.8 months (95% CI 2.3-10.1). Median time to progression is 6 months (95% CI 3.0-NA). UVA showed worse OS is associated with higher neutrophillymphocyte ratio (NLR) (HR 1.1; 95% CI 1.0-1.1, p<0.01), higher bone marrow SUV mean (BMMean) (HR 7.5; 95% CI 2.3-24.7, p<0.01), and higher bone marrow-liver SUV ratio (BLR) (HR 9.2; 95% CI 1.2-68.3, p Z 0.03). Disease progression was more likely with higher NLR (HR 1.07; 95% CI 1.02-1.12, p<0.01), and higher BMMean (HR 5.72; 95% CI 1.91-17.1, p<0.01). UVA revealed no significant association between PFS or OS and T-stage at diagnosis, performance status, p16 status, tobacco use, and quantitative imaging features of the liver and spleen. On MVA, NLR and BMMean were predictive for OS (HR 1.1; 95% CI 1.0-1.1, p<0.01 and HR 4.8; 95% CI 1.5-15.4, p<0.01, respectively) which was also true for PFS (HR 1.06; 95% CI 1.02-1.10, p<0.01 and HR 4.47; 95% CI 1.53-13.1 p<0.01, respectively). Conclusion: This exploratory analysis identified bone marrow SUV and NLR as important predictors for response to immunotherapy in R/M HNC. Given the limited data maturity, longer follow up in a larger cohort of patients is needed to verify these results. These results suggest that the combination of novel pre-treatment biomarkers may help guide future efforts for treatment intensification in poor-prognosis R/M HNC patients.

Prospective validation of diffusion-weighted MRI as a biomarker of tumor response and oncologic outcomes in head and neck cancer: Results from an observational biomarker pre-qualification study

“…The selection of these clinical variables was based on the findings of our group’s large-scale HNC clinical models’ performance for survival endpoints prediction. (31) Subsequently, additive models that include the clinical parameters plus ΔADC were constructed to assess the potential additive value of the imaging parameter. Models were only constructed for patients with a GTV-P.…”

Construction of Graduate Behavior Dynamic Model Based on Dynamic Decision Tree Algorithm

“…Based on this result, it has good social significance for college graduates to change their strategies according to the current learning situation and improve their school efficiency [11]. Dijk et al proposed to establish the largest information database for college graduates, which aims to use big data technology to predict the whereabouts of college graduates, so as to improve the statistical efficiency of the whereabouts of graduates [12]. Xiao et al analyzed the data of the ship automatic identification system by using adaptive learning, motion modeling, and particle filter technology through big data acquisition and aiming at the direction of intelligent maritime traffic and carried out the collision risk assessment of ship navigation dynamic trajectory.…”